MoCha-Stereo 抹茶算法

[CVPR2024] The official implementation of "MoCha-Stereo: Motif Channel Attention Network for Stereo Matching".

[Arxiv] The extension version of MoCha-Stereo. "Motif Channel Opened in a White-Box: Stereo Matching via Motif Correlation Graph"

Demo.mp4

V1 Version

     

MoCha-Stereo: Motif Channel Attention Network for Stereo Matching
Ziyang Chen†, Wei Long†, He Yao†, Yongjun Zhang✱,Bingshu Wang, Yongbin Qin, Jia Wu
CVPR 2024
Contact us: [email protected]; [email protected]✱

@inproceedings{chen2024mocha,
  title={MoCha-Stereo: Motif Channel Attention Network for Stereo Matching},
  author={Chen, Ziyang and Long, Wei and Yao, He and Zhang, Yongjun and Wang, Bingshu and Qin, Yongbin and Wu, Jia},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={27768--27777},
  year={2024}
}

V2 Version

 

Motif Channel Opened in a White-Box: Stereo Matching via Motif Correlation Graph
Ziyang Chen, Yongjun Zhang✱,Wenting Li, Bingshu Wang, Yong Zhao, C. L. Philip Chen
Arxiv Report
Contact us: [email protected]; [email protected]✱

FAQ

Q1: Weight for "tf_efficientnetv2_l"? (Please refer to issue #6 "关于tf_efficientnetv2_l检查点的问题", #8 "预训练权重", and #9 "code error". )

A1: You can download it here. Moreover, the weight pretrained on the Scene Flow dataset is available here.

Q2: How to visualize the disparity map? (Please refer to issue #15 "询问可视化问题", and #10 "请问这个项目如何可视化推理的结果呢")

A2: You can accomplish this using "demo.py".

python demo.py --restore_ckpt ./model/mocha-stereo.pth -l ./datasets/Middlebury/MiddEval3/trainingF/*/im0.png -r ./datasets/Middlebury/MiddEval3/trainingF/*/im1.png --output_directory ./your/path

The libary "matplotlib" is required for visualizing the disparity map.

Todo List

• [CVPR2024] V1 version
  • Paper
  • Code of MoCha-Stereo
• V2 version
  • Preprint manuscript
  • Code of MoCha-V2

Acknowledgements

• This project borrows the code from IGEV, DLNR, RAFT-Stereo, GwcNet. We thank the original authors for their excellent works!
• Grateful to Prof. Wenting Li, Prof. Huamin Qu, Dr. Junda Cheng, Mr./Mrs. "DLUTTengYH", Mr./Mrs. "YHCks", and anonymous reviewers for their comments on "MoCha-Stereo: Motif Channel Attention Network for Stereo Matching" (V1 version of MoCha-Stereo).